meta-llama Open weights Vision

Llama 4 Maverick pricing & benchmarks

Released Apr 5, 2025 · served by 5 providers

Input / 1M tokens$0.200
Output / 1M tokens$0.800
Context window1.0M
Cached input
Intelligence index14.3
Coding index16.3
Agentic index1.3
1M in + 300K out$0.44

What Llama 4 Maverick is

Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...

Providers and prices for Llama 4 Maverick

The same model costs different amounts depending on who serves it. Prices are per 1M tokens, cheapest first. Uptime and throughput are OpenRouter's rolling measurements, not vendor claims.

ProviderInputOutputContext QuantUptime 24hThroughput
DeepInfracheapest $0.200 $0.800 1.0M fp8 99.8%
Novita $0.270 $0.850 1.0M fp8 99.8%
DigitalOcean $0.250 $0.870 128K 99.8%
Parasail $0.350 $1.00 524K fp8 99.9%
Google $0.350 $1.15 524K 99.9%

Llama 4 Maverick benchmark results

Design Arena head-to-head results, as reported through the OpenRouter model API.

CategoryArenaRankEloWin rate
3d models #99 957 40.2%
uicomponent models #103 936 40.8%
dataviz models #108 912 38.4%
gamedev models #109 894 33.7%
codecategories models #111 910 35.8%
website models #114 896 34.4%

Cheaper models in the same class

Models scoring within 4 points of Llama 4 Maverick on the Intelligence Index, but with a lower output price.

FAQ

How much does the Llama 4 Maverick API cost?

$0.200 per 1M input tokens and $0.800 per 1M output tokens. A typical workload of 1M input + 300K output tokens costs about $0.44.

Is Llama 4 Maverick free?

No. It is a paid model starting at $0.200 per 1M input tokens, though some providers offer trial credits.

Which provider is cheapest for Llama 4 Maverick?

DeepInfra at $0.200 input / $0.800 output per 1M tokens (fp8 quantization).

Can I self-host Llama 4 Maverick?

Yes — weights are published as meta-llama/Llama-4-Maverick-17B-128E-Instruct on Hugging Face, so you can run it on your own hardware.